Spinal deformity is a disease that causes a three-dimensional deformation of the spinal column. When it worsens, surgery is required to screw correction rods to the spinal column. However, the surgery requires intraoperative rod bending work, which burdens the patients and causes unexpected rod breakage inside the body. Therefore, "pre-bent" rods comprising several rods with standardized shapes have been proposed to solve these problems. When designing pre-bent rods, knowing the number of rods to be prepared and the kinds of shapes required is essential. In this paper, we propose a geometric processing technique to identify an optimal set of these standardized pre-bent rod shapes for surgeries on adult spinal deformity and describe the similarity evaluation among existing rod shapes using CT scan, medial axis extraction, and iterative closest point algorithm. Moreover, we present the derivation of standardized rod shapes using hierarchical cluster analysis and the best fit of the B-spline curve to each cluster. Finally, we discuss the effectiveness of prebent rod shapes derived from CT scans of 26 existing rods of 13 patients.

No abstract